Animation retargeting

Abstract

Spatial semantics detection is performed on an animation transfer source model including N first local object BVs, to obtain a spatial semantics detection result. Based on the spatial semantics detection result and from the N first local object BVs of the first object, effective interaction BVs including a pair of first local object BVs that satisfies an animation retargeting policy are determined. From an animation transfer target model including M second local object BVs, to-be-transferred BVs that satisfy the animation retargeting policy are determined. Spatial semantics information of the effective interaction BVs is transferred to the to-be-transferred BVs, to obtain transferred BVs that carry the spatial semantics information. A model correction parameter is acquired based on the spatial semantics information. Based on the model correction parameter, model correction processing is performed on the animation transfer target model that comprises the transferred BVs, to obtain an updated animation transfer target model.

What is claimed is: 
     
         1 . A method of animation retargeting, the method comprising:
 displaying, on an animation retargeting configuration interface, an animation transfer source model associated with a first object, and an animation transfer target model associated with a second object, the first object comprising at least N first local objects, the second object comprising at least M second local objects, N and M being positive integers, the animation transfer source model comprising N first local object bounding volumes (BVs) respectively associated with the N first local objects of the first object, and the animation transfer target model comprising M second local object BVs respectively associated with the M second local objects of the second object;   performing spatial semantics detection on the N first local object BVs, to obtain a spatial semantics detection result;   determining, based on the spatial semantics detection result and from the N first local object BVs of the first object, effective interaction BVs including a pair of first local object BVs that satisfies an animation retargeting policy;   determining, from the M second local object BVs of the second object, to-be-transferred BVs that satisfy the animation retargeting policy;   transferring spatial semantics information of the effective interaction BVs to the to-be-transferred BVs to obtain transferred BVs that carry the spatial semantics information;   acquiring a model correction parameter based on the spatial semantics information;   performing, based on the model correction parameter, model correction processing on the animation transfer target model that comprises the transferred BVs, to obtain an updated animation transfer target model; and   displaying the updated animation transfer target model on the animation retargeting configuration interface.   
     
     
         2 . The method according to  claim 1 , wherein:
 the performing the spatial semantics detection comprises:
 acquiring, from the N first local object BVs, a first BV and a second BV that satisfy a local contact condition defined in a contactable definition table associated with the N first local object BVs; 
 determining a first geometry attribute of the first BV and a second geometry attribute of the second BV; 
 determining, based on the first geometry attribute, the second geometry attribute, and a contact relationship that is indicated by the local contact condition, a first local contact surface of the first BV and a second local contact surface of the second BV; 
 determining, based on the first local contact surface and the second local contact surface, a first closest point pair between the first BV and the second BV, wherein the first closest point pair comprises a first closest point on the first BV and a second closest point on the second BV; and 
 determining a closest point distance between the first closest point and the second closest point; and 
   the spatial semantics detection result includes spatial semantics of the first BV and the second BV including the first closest point pair and the closest point distance.   
     
     
         3 . The method according to  claim 2 , wherein:
 the first geometry attribute and the second geometry attribute are of a sphere attribute; and   the determining the closest point distance comprises:
 determining a spherical center distance between a first spherical center of the first BV and a second spherical center of the second BV; and 
 determining the closest point distance between the first closest point and the second closest point according to the spherical center distance, a first radius of the first BV, and a second radius of the second BV. 
   
     
     
         4 . The method according to  claim 2 , wherein:
 the first geometry attribute is of a sphere attribute, and the second geometry attribute is of a capsule attribute; and   the determining the closest point distance comprises:
 determining, based on a spherical center distance between a first spherical center of the first BV and a second spherical center of the second BV, a first radius of the first BV, and a second radius of the second BV, the closest point distance between the first closest point and the second closest point when the contact relationship indicates that the first BV is located on a side surface of a sphere portion of the second BV. 
   
     
     
         5 . The method according to  claim 2 , wherein:
 the first geometry attribute is of a sphere attribute, the second geometry attribute is of a capsule attribute; and   the determining the closest point distance comprises:
 determining, based on a spherical center distance between a spherical center of the first BV and a cylinder center line segment of the second BV, a first radius of the first BV, and a second radius of the second BV, the closest point distance between the first closest point and the second closest point when the contact relationship indicates that the first BV is located on a side surface of a cylinder portion of the second BV. 
   
     
     
         6 . The method according to  claim 2 , wherein:
 the first geometry attribute and the second geometry attribute are of a capsule attribute, the first BV comprises a first cylinder and a first sphere, the second BV comprises a second cylinder and a second sphere; and   the determining the closest point distance comprises:
 determining, when the contact relationship indicates that a first cylinder center line segment of the first BV is not located on a side surface of the second cylinder of the second BV and a second cylinder center line segment of the second BV is not located on a side surface of the first cylinder of the first BV, a closest endpoint distance between the first cylinder center line segment of the first BV and the second cylinder center line segment of the second BV; and 
 determining, based on the closest endpoint distance, a first radius of the first BV, and a second radius of the second BV, the closest point distance between the first closest point and the second closest point. 
   
     
     
         7 . The method according to  claim 2 , wherein:
 the first geometry attribute and the second geometry attribute are of a capsule attribute, the first BV comprises a first cylinder and a first sphere, the second BV comprises a second cylinder and a second sphere; and   the determining the closest point distance comprises:
 determining, when the contact relationship indicates that a first target endpoint of a first cylinder center line segment of the first BV is located on a side surface of the second cylinder of the second BV and a second target endpoint of a second cylinder center line segment of the second BV is located on a side surface of the first cylinder of the first BV, a target distance between the first target endpoint of the first cylinder center line segment of the first BV and the second target endpoint of the second cylinder center line segment of the second BV; and 
 determining, based on the target distance, a first radius of the first BV, and a second radius of the second BV, the closest point distance between the first closest point and the second closest point. 
   
     
     
         8 . The method according to  claim 2 , wherein:
 the first geometry attribute and the second geometry attribute are of a capsule attribute; and   the determining the closest point distance comprises:
 determining, when the contact relationship indicates that the first closest point is projected on a first cylinder center line segment of the first BV and second closest point is projected on a second cylinder center line segment of the second BV, that a first distance from the first closest point to the first cylinder center line segment of the first BV is a first radius of the first BV and that a second distance from the second closest point to the second cylinder center line segment of the second BV is a second radius of the second BV; 
 determining a second closest point distance between the first cylinder center line segment of the first BV and the second cylinder center line segment of the second BV; and 
 determining, based on the second closest point distance, the first radius of the first BV, and the second radius of the second BV, the closest point distance between the first closest point and the second closest point. 
   
     
     
         9 . The method according to  claim 1 , wherein:
 the animation retargeting policy comprises an interaction detection policy, and the interaction detection policy comprises an interaction distance threshold for performing interaction detection; and   the method further comprises:
 acquiring, from the N first local object BVs and based on the spatial semantics detection result, a first BV and a second BV; 
 determining a first closest point distance between the first BV and the second BV; and 
 determining, when the first closest point distance between the first BV and the second BV is less than the interaction distance threshold, the first BV and the second BV as the pair of first local object BVs. 
   
     
     
         10 . The method according to  claim 1 , wherein:
 the effective interaction BVs comprise a first effective interaction BV and a second effective interaction BV, and the to-be-transferred BVs comprise a first to-be-transferred BV and a second to-be-transferred BV;   the spatial semantics information comprises a closest point pair between the first effective interaction BV and the second effective interaction BV, first spatial semantics information of the first effective interaction BV, and second spatial semantics information of the second effective interaction BV;   the transferred BVs comprise a first transferred BV and a second transferred BV;   the closest point pair between the first effective interaction BV and the second effective interaction BV comprises a first closest point of the first effective interaction BV and a second closest point of the second effective interaction BV; and   the transferring the spatial semantics information comprises:
 determining a first position point on the first to-be-transferred BV to correspond to the first closest point of the first effective interaction BV; 
 determining a second position point on the second to-be-transferred BV to correspond to the second closest point of the second effective interaction BV; 
 transferring the first spatial semantics information of the first effective interaction BV to the first to-be-transferred BV based on the first position point, to obtain the first transferred BV carrying the first spatial semantics information; 
 transferring the second spatial semantics information of the second effective interaction BV to the second to-be-transferred BV based on the second position point, to obtain the second transferred BV carrying the second spatial semantics information; and 
 generating, based on the first transferred BV carrying the first spatial semantics information and the second transferred BV carrying the second spatial semantics information, the transferred BVs carrying the spatial semantics information. 
   
     
     
         11 . The method according to  claim 10 , further comprising:
 constructing a first coordinate system based on the first effective interaction BV;   determining a local vector of the first closest point of the first effective interaction BV, the local vector pointing from an origin of the first coordinate system to the first closest point; and   generating, based on the local vector of the first closest point of the first effective interaction BV, the first spatial semantics information of the first effective interaction BV.   
     
     
         12 . The method according to  claim 11 , wherein:
 the first effective interaction BV is a capsule; and   the generating the first spatial semantics information of the first effective interaction BV comprises:
 determining a proportion coefficient corresponding to the first closest point of the first effective interaction BV on an axial vector of the first effective interaction BV; 
 projecting the local vector, to obtain a projection vector of the first closest point of the first effective interaction BV; and 
 generating, based on the proportion coefficient and the projection vector, the first spatial semantics information of the first effective interaction BV. 
   
     
     
         13 . The method according to  claim 10 , wherein the acquiring the model correction parameter comprises:
 determining, based on the animation transfer target model comprising the transferred BVs in the animation retargeting configuration interface, that the first transferred BV corresponds to an End-effector (EEF);   forming a second closest point pair between the first transferred BV and the second transferred BV, the second closest point pair including the first position point and the second position point;   superposing, on the second position point in the second closest point pair, a closest point distance between the first effective interaction BV and the second effective interaction BV along a direction perpendicular to a surface of the first transferred BV, to obtain an interaction estimation point;   determining, based on a difference between the interaction estimation point and the first position point, and an adjustment weight, an adjustment vector of the first transferred BV; and   generating the model correction parameter based on the adjustment vector of the first transferred BV.   
     
     
         14 . The method according to  claim 13 , wherein the performing the model correction processing comprises:
 performing, based on the adjustment vector of the first transferred BV, a translation adjustment on the first transferred BV in the animation transfer target model, to obtain an updated first transferred BV; and   performing, based on the updated first transferred BV, the model correction processing on the animation transfer target model, to obtain the updated animation transfer target model.   
     
     
         15 . The method according to  claim 1 , wherein:
 the animation retargeting configuration interface comprises an animation transfer source model option and an animation transfer target model option; and   the method further comprises:
 determining, in response to a trigger operation on the animation transfer source model option, a target frame of animation data from a plurality of frames of animation data in target animation data, the target frame of animation data including the animation transfer source model; and 
 selecting, in response to a trigger operation on the animation transfer target model option, the animation transfer target model from a transfer model database. 
   
     
     
         16 . The method according to  claim 1 , wherein:
 the animation retargeting configuration interface comprises a first BV initialization option and a second BV initialization option; and   the displaying the animation transfer source model and the animation transfer target model comprises:
 displaying, in response to a trigger operation on the first BV initialization option, N first initial local object BVs associated with the first object; 
 unbinding the N first initial local object BVs from the animation transfer source model, to obtain unbound N first initial local object BVs; 
 adjusting, in response to an adjustment operation on the unbound N first initial local object BVs, the unbound N first initial local object BVs to obtain the N first local object BVs associated with the first object; 
 displaying the N first local object BVs associated with the first object on the animation retargeting configuration interface; 
 displaying, in response to a trigger operation on the second BV initialization option, M second initial local object BVs associated with the second object; 
 adjusting, in response to an adjustment operation on the M second initial local object BVs, the M second initial local object BVs to obtain the M second local object BVs associated with the second object; and 
 displaying the M second local object BVs associated with the second object on the animation retargeting configuration interface. 
   
     
     
         17 . An information processing apparatus for animation retargeting, comprising processing circuitry configured to:
 display, on an animation retargeting configuration interface, an animation transfer source model associated with a first object, and an animation transfer target model associated with a second object, the first object comprising at least N first local objects, the second object comprising at least M second local objects, N and M being positive integers, the animation transfer source model comprising N first local object bounding volumes (BVs) respectively associated with the N first local objects of the first object, and the animation transfer target model comprising M second local object BVs respectively associated with the M second local objects of the second object;   perform spatial semantics detection on the N first local object BVs, to obtain a spatial semantics detection result;   determine, based on the spatial semantics detection result and from the N first local object BVs of the first object, effective interaction BVs including a pair of first local object BVs that satisfies an animation retargeting policy;   determine, from the M second local object BVs of the second object, to-be-transferred BVs that satisfy the animation retargeting policy;   transfer spatial semantics information of the effective interaction BVs to the to-be-transferred BVs to obtain transferred BVs that carry the spatial semantics information;   acquire a model correction parameter based on the spatial semantics information;   perform, based on the model correction parameter, model correction processing on the animation transfer target model that comprises the transferred BVs, to obtain an updated animation transfer target model; and   display the updated animation transfer target model on the animation retargeting configuration interface.   
     
     
         18 . The information processing apparatus according to  claim 17 , wherein
 the processing circuitry is configured to:
 acquire, from the N first local object BVs, a first BV and a second BV that satisfy a local contact condition defined in a contactable definition table associated with the N first local object BVs; 
 determine a first geometry attribute of the first BV and a second geometry attribute of the second BV; 
 determine, based on the first geometry attribute, the second geometry attribute, and a contact relationship that is indicated by the local contact condition, a first local contact surface of the first BV and a second local contact surface of the second BV; 
 determine, based on the first local contact surface and the second local contact surface, a first closest point pair between the first BV and the second BV, wherein the first closest point pair comprises a first closest point on the first BV and a second closest point on the second BV; and 
 determine a closest point distance between the first closest point and the second closest point; and 
   the spatial semantics detection result includes spatial semantics of the first BV and the second BV including the first closest point pair and the closest point distance.   
     
     
         19 . The information processing apparatus according to  claim 18 , wherein:
 the first geometry attribute and the second geometry attribute are of a sphere attribute; and   the processing circuitry is configured to:
 determine a spherical center distance between a first spherical center of the first BV and a second spherical center of the second BV; and 
 determine the closest point distance between the first closest point and the second closest point according to the spherical center distance, a first radius of the first BV, and a second radius of the second BV. 
   
     
     
         20 . A non-transitory computer-readable storage medium storing instructions which when executed by at least one processor cause the at least one processor to perform:
 displaying, on an animation retargeting configuration interface, an animation transfer source model associated with a first object, and an animation transfer target model associated with a second object, the first object comprising at least N first local objects, the second object comprising at least M second local objects, N and M being positive integers, the animation transfer source model comprising N first local object bounding volumes (BVs) respectively associated with the N first local objects of the first object, and the animation transfer target model comprising M second local object BVs respectively associated with the M second local objects of the second object;   performing spatial semantics detection on the N first local object BVs, to obtain a spatial semantics detection result;   determining, based on the spatial semantics detection result and from the N first local object BVs of the first object, effective interaction BVs including a pair of first local object BVs that satisfies an animation retargeting policy;   determining, from the M second local object BVs of the second object, to-be-transferred BVs that satisfy the animation retargeting policy;   transferring spatial semantics information of the effective interaction BVs to the to-be-transferred BVs to obtain transferred BVs that carry the spatial semantics information;   acquiring a model correction parameter based on the spatial semantics information;   performing, based on the model correction parameter, model correction processing on the animation transfer target model that comprises the transferred BVs, to obtain an updated animation transfer target model; and   displaying the updated animation transfer target model on the animation retargeting configuration interface.